Xijie Huang (Owen) 黄悉偈

I am a PhD Student at Hong Kong University of Science and Technology (HKUST), where I work on efficient deep learning. I am a member of Vision and System Design Lab , advised by Prof. Tim Kwang-Ting CHENG. I work closely with Zechun Liu from Meta Reality Lab. Currently, I am working as a research intern at Snap Research Creative Vision Group, supervised by Jian Ren and Anil Kag. I have also conducted a research internship at Microsoft Research Asia (MSRA). Previously, I work with Prof. Zhiqiang Shen from MBZUAI, Prof. Cewu Lu from SJTU, and Prof. Mani Srivastava from UCLA.

I am seeking both industrial and academic positions starting Fall 2025. Feel free to reach out via email (huangxijie1108 AT gmail.com) if there are opportunities aligned with my expertise.

CV  /  Google Scholar  /  Twitter (X)  /  Zhihu (知乎)  /  Github  /  Openreview

[2024-09] Two papers (CoT-Influx and RoLoRA) are accepted by EMNLP 2024

[2024-08] Two papers (ACS-QAT and Quantization Variation) are accepted by TMLR

[2024-07] Start my intership at Snap Research Santa Monica

[2023-09] One paper (LLM-FP4) is accepted by EMNLP 2023

[2023-05] Start my internship at Microsoft Research Asia (MSRA)

[2022-05] One paper (SDQ) is accepted by ICML 2022

[2021-01] One paper (TIN) is accepted by TPAMI

[2020-08] Start my PhD at Hong Kong University of Science and Technology (HKUST)

[2020-06] Graduate from Shanghai Jiao Tong University (SJTU)

[2020-03] One paper (PaStaNet) is accepted by CVPR 2020

[2019-06] Start my research internship at University of California, Los Angeles (UCLA)

[2019-03] One paper (TIN) is accepted by CVPR 2019

My research interests lie in the general area of artificial intelligence, particularly in efficient deep learning and human-centric computer vision. More concretely, My research interests focus on quantization, algorithm-hardware co-design, human-object interation, and healthcare.

Efficient AI Algorithm

A LoRA-based scheme for effective weight-activation quantization. RoLoRA utilizes rotation for outlier elimination and proposes rotation-aware fine-tuning to preserve the outlier-free characteristics in rotated LLMs.

A novel approach to push the boundaries of few-shot CoT learning to improve LLM math reasoning capabilities. We propose a coarse-to-fine pruner as a plug-and-play module for LLMs, which first identifies crucial CoT examples from a large batch and then further prunes unimportant tokens.

A new angle through the coreset selection to improve the training efficiency of quantization-aware training. Our method can achieve an accuracy of 68.39% of 4-bit quantized ResNet-18 on the ImageNet-1K dataset with only a 10% subset, which has an absolute gain of 4.24% compared to the previous SoTA.

An analysis of the underlying difficulty of transformer quantization in the view of variation. A multi-crop knowledge distillation-based quantization method is proposed.

A per-channel activation quantization scheme with additional scaling factors that can be reparameterized as exponential biases of weights, incurring a negligible cost. Our method can quantize both weights and activations in the Bert model to only 4-bit and achieves an average GLUE score of 80.07, which is only 3.66 lower than the full-precision model, significantly outperforming the previous state-of-the-art method that had a gap of 11.48.

A novel stochastic quantization framework to learn the optimal mixed precision quantization strategy.

Efficient AI Hardware

A hybrid attention mechanism coupled with a KV-weight-reused scheduler to fuse the layers between convolution and attention. Additionally, we introduce a cascaded feature map pruning strategy to achieve unified convolution-attention pruning.

A genetic LUT-Approximation algorithm namely GQA-LUT that can automatically determine the parameters with quantization awareness. The results demonstrate that GQA-LUT achieves negligible degradation on the challenging semantic segmentation task for both vanilla and linear Transformer models.

A tiny accelerator for mixed-bit sparse CNNs featuring a novel scheme of single vector-based compressed sparse filter (CSF) method and single input multiple output scratch pad (SIMO SPad) to effectively compress weight and fetch the needed input activation.

Human-centric Vision

A visual analytics approach to support flexible and scalable video data programming for model steering with reduced human effort.

An automatic, vision-based system for monitoring and analyzing the physical and mental well-being of senior citizens.

A label-agnostic unified federated learning framework, named FedMix, for medical image segmentation based on mixed image labels. In FedMix, each client updates the federated model by integrating and effectively making use of all available labeled data ranging from strong pixel-level labels, weak bounding box labels, to weakest image-level class labels.

A transferable knowledge learner and can be cooperated with any HOI detection models to achieve desirable results. TIN outperforms state-of-the-art HOI detection results by a great margin, verifying its efficacy and flexibility.

A large-scale knowledge base PaStaNet, which contains 7M+ PaSta annotations. And two corresponding models are proposed: first, we design a model named Activity2Vec to extract PaSta features, which aim to be general representations for various activities.

A latent fingerprint enhancement method based on the progressive generative adversarial network (GAN).

A large-scale Human Activity Knowledge Engine (HAKE) based on the human body part states to promote the activity understanding.

A framework to detect trojan backdoors in DNNs via output explanation techniques

Thanks to Barron's website template .